On linear accuracy estimates of Tikhonov's method

We investigate the rate of convergence of Tikhonov's scheme for solving irregular nonlinear equations with smooth operators in a Hilbert space in assumption that the derivative of the operator at the solution is normally solvable. With an appropriate a priori and a posteriori coordination of the regularization parameter and the level of errors in input data, we prove that the accuracy estimate is proportional to the error level. Without using the normal solvability condition, we establish similar estimates for the convergence rate in proper subspaces of the symmetrized derivative at the solution and at the current Tikhonov's approximation.